Underwater three-dimensional reconstruction method and system based on UUV cruise and two-dimensional sonar

Abstract

The invention discloses an underwater three-dimensional reconstruction method and system based on UUV cruise and two-dimensional sonar. The system comprises a UUV carrying platform, an environment sensing module, a data processing module and a three-dimensional reconstruction method. Wherein the UUV carrying platform undertakes simple motion and control functions, the environment sensing module performs data acquisition through the three-dimensional reconstruction method, and the data processing module operates a three-dimensional reconstruction algorithm after receiving the acquired data to realize three-dimensional reconstruction of an underwater scene. According to the underwater acoustic three-dimensional reconstruction method, the two-dimensional sonar is carried by the UUV, three-dimensional reconstruction of an underwater structure can be achieved with low cost, high efficiency and high precision, and environment perception guarantee is provided for underwater robot operation and maintenance in a turbid water area; and the method can also be directly applied to underwater structure defect detection of highly turbid water areas, such as cavities and large-area peeling.

Description

Technical Field

[0001] The present invention relates to the field of underwater environment perception, and in particular to an underwater three-dimensional reconstruction method and system based on UUV cruising and two-dimensional sonar. Background Art

[0002] The three-dimensional reconstruction of underwater structures (such as offshore platform jacket, wind power platform pile foundation, bridge pier, dam, etc.) is a key technology for underwater robots in target search, mapping, and operation. Among them, the three-dimensional reconstruction in turbid water is a difficult problem in underwater three-dimensional reconstruction. Since the signal attenuation of optical sensors is severe in turbid waters, three-dimensional reconstruction methods such as cameras and lasers all fail, and it can only be achieved through three-dimensional acoustic devices.

[0003] Conventional three-dimensional sonar devices (such as BlueView BV5000 three-dimensional sonar, Echoscope three-dimensional sonar of the British CodaOctopus company, etc.) are expensive and large in volume, and carry data processing modules of the same large volume. Generally, they need to be towed by a ship for imaging or deployed manually for imaging, which results in inflexible imaging and low efficiency. At the same time, due to their large volume, it is also difficult to be carried on small and medium-sized underwater robots, resulting in the inability to be applied to three-dimensional reconstruction in deeper waters and automated three-dimensional reconstruction of robots underwater. Here, a method for high-precision three-dimensional reconstruction based on UUV cruising and carrying two-dimensional image sonar is proposed.

[0004] The UUV realizes depth-keeping control at a certain depth through a depth gauge, which is relatively easy and has good results in waters with non-high-speed water flows; the 2D SLAM method based on lidar has been widely used on land robots. Combining these two technologies, at a fixed depth, the sonar image is processed into point cloud data similar to lidar, and using the 2D SLAM algorithm, it is relatively easy to obtain the two-dimensional map and robot pose information at this depth.

[0005] On the basis of obtaining the two-dimensional map at a certain depth, if the control algorithm is excellent enough to control it to remain stationary in the x, y azimuth and heading angle, after obtaining the x and y coordinates in the robot reference plane, only moving in the z-axis direction can obtain the three-dimensional coordinate position of the robot underwater, and at the same time perform two-dimensional scanning of the environment to obtain three-dimensional environment information. This is obviously too ideal. Since the environmental information except the two-dimensional map of the original depth plane is unknown, the robot lacks an absolute positioning reference, and the positioning and attitude errors will inevitably increase with time. Summary of the Invention

[0006] The purpose of the present invention is to propose an underwater three-dimensional reconstruction method and system based on UUV cruising and two-dimensional sonar in view of the deficiencies of the prior art.

[0007] The object of the present invention is achieved by the following technical solutions: An underwater three-dimensional reconstruction method based on UUV cruise and two-dimensional image sonar, the method comprising:

[0008] S1. Obtain a first target depth and a second target depth according to the height range of the space to be reconstructed, and lower the robot to the first target depth and the second target depth for depth fixation;

[0009] S2. Use the robot to collect dead reckoning information and forward-looking sonar information, and utilize the obtained information to construct a target depth plane map to obtain a first target depth plane map and a second target depth plane map;

[0010] S3. Move the robot to the first target depth for positioning and move and scan along the vertical direction; when the robot moves to the second target depth, perform secondary positioning and correct the movement route according to the coordinate difference in the map;

[0011] S4. Based on the corrected path information, splice and filter the point cloud in the forward-looking sonar information to obtain spatial point cloud information;

[0012] S5. Obtain the spatial point cloud information scanned by different paths for registration to obtain the overall three-dimensional point cloud reconstruction of the underwater structure.

[0013] Further, the dead reckoning information is used to calculate the current position and attitude of the robot, specifically including: the three-axis velocity obtained by the DVL, the three-axis attitude, angular velocity, acceleration obtained by the IMU, and the depth obtained by the depth gauge.

[0014] Further, the forward-looking sonar information is the occupancy probability of the spatial point cloud converted from the forward-looking sonar image, and the conversion process is specifically as follows:

[0015] Convert the image with length and width of R pixel obtained by the sonar into a spatial voxel with the number of length, width and height being R pixel , respectively; where R pixel represents the number of pixels of the sonar imaging radius on the image, and each voxel is represented by a uniformly distributed point cloud. Taking the sonar imaging origin as the origin of the spatial coordinate system, the default value of each point is -1, indicating unknown;

[0016] Traverse each point cloud (X, Y, Z) and convert it to the polar coordinate system

[0017]

[0018] For those satisfying R ∈ [0, R pixel , The points are the actual imaging area corresponding to the sonar image of this frame. Through the corresponding R and θ, the occupancy probability of this voxel is obtained from the sonar image, where θ sonar is the azimuth angle of the sonar, and sonar is the vertical aperture of the sonar

[0019] The occupancy probability of the point cloud (X, Y, Z) is normalized, that is

[0020]

[0021] I max is the maximum pixel value in the sonar image, and I min is the minimum pixel value. The range of P(X, Y, Z) is (0 to 1). The larger the value, the greater the probability that the voxel space corresponding to this point cloud is occupied.

[0022] Furthermore, the construction of the target depth plane map includes: using the dead reckoning result in this depth plane as the odometer input, and the leading point cloud information of the forward-looking sonar image as the point cloud input, and completing the construction using the two-dimensional cartographer algorithm.

[0023] Furthermore, the correction of the motion route includes: The matching coordinates of the robot in the first target depth plane map are (x 1 , y 1 ), the coordinates of the robot in the map of the second target depth plane are (x 2 , y 2 ), and the position coordinates (x(d i ), y(d i )) at depth d i are corrected to

[0024]

[0025] where d 1 < d i < d 2 .

[0026] Furthermore, the splicing and filtering of the point cloud in the forward-looking sonar information includes: For each voxel, keep the minimum value among all observations. The specific steps are as follows:

[0027] S6.1 Add the point P new generated by the latest frame to the environmental point cloud P e and update the value of the original environmental point cloud P e

[0028]

[0029] S6.2: After completing the survey line, process the global point cloud on this survey line and set an intensity threshold E threshold , and remove the points in the point cloud of this survey line that are less than this energy threshold, that is, remove the uncertain points and low-probability points. The remaining points are the three-dimensional imaging results on this survey line.

[0030] Furthermore, for the spatial point cloud information scanned in different paths, when selecting to switch the survey line on the depth plane where the target depth plane map is constructed, the positioning information of the ROV during the survey line switch can be enhanced based on the target depth plane map.

[0031] According to another aspect of the specification, the present specification also provides a system for implementing the above method. This system includes: a UUV carrier platform, an environmental perception module, and a data processing module;

[0032] The UUV carrier platform is an ROV or an AUV, which is used to undertake the motion and control functions, and is equipped with an environmental perception module to collect dead reckoning information and forward-looking sonar information; after receiving the collected data, the data processing module realizes the three-dimensional reconstruction of the underwater scene;

[0033] The environmental perception module includes a motion perception module and an acoustic perception module; the motion perception module is an IMU, DVL, depth gauge; the acoustic perception module is a forward-looking sonar and an altimeter sonar.

[0034] Furthermore, the UUV carrier platform includes a mechanical body, an actuator, and a motion controller; the mechanical body includes a frame and buoyancy materials; the actuator includes thrusters and a pan-tilt; the motion controller is used to receive the control instructions issued by the data processing module and complete the control of the actuator.

[0035] According to another aspect of the specification, the present specification also provides an underwater three-dimensional reconstruction device based on UUV cruising and two-dimensional image sonar, including a memory and one or more processors. Executable code is stored in the memory, and when the processor executes the executable code, the above-mentioned underwater three-dimensional reconstruction method based on UUV cruising and two-dimensional image sonar is realized.

[0036] Advantages of the present invention:

[0037] 1. In this method, the UUV carries a multi-beam sonar for scanning and imaging, stitches the frontiers of sonar images based on the motion trajectory, and corrects the UUV path through prior information and two-dimensional map matching, making the scanning trajectory more accurate. Compared with using a three-dimensional sonar, the cost is lower; compared with ship-towed imaging or manual deployment imaging, the imaging flexibility and efficiency are higher.

[0038] 2. This method is based on the idea of ​​synthetic aperture and space carving, and processes vertical scanning sonar data to reduce the uncertainty in vertical aperture. When targeting underwater vertical structures such as bridge piers, dam bodies, wind turbine pile foundations, and marine conductor racks, the imaging plane is nearly perpendicular to the surface of the structure during the imaging process, which has higher imaging accuracy than the 3D reconstruction method based on pan-tilt scanning (BlueView BV5000 3D sonar).

[0039] 3. The present invention only needs to deploy conventional PID control and Kalman filtering algorithms on the robot motion controller, with low computational cost. The environmental perception data is transmitted to the dry-end computer through the network to run the three-dimensional reconstruction algorithm, making the system easy to deploy and can be well deployed in small underwater robots.

[0040] 4. Based on dead reckoning and depth plane maps, the robot can know its specific position. Combined with motion control methods, it can first simply explore the underwater environment to avoid accidents during the subsequent three-dimensional reconstruction process, and can operate in waters with complex water conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a system block diagram of the underwater three-dimensional reconstruction system based on UUV cruising and two-dimensional sonar of the present invention;

[0042] Figure 2 An axonometric diagram of the operation process of an embodiment of the present invention;

[0043] Figure 3 It is a left side view of the operation process of one embodiment of the present invention;

[0044] Figure 4 The d1 depth plane map obtained by an embodiment of the present invention;

[0045] Figure 5 The d2 depth plane map obtained by an embodiment of the present invention;

[0046] Figure 6 A schematic diagram of a route of an operation process according to an embodiment of the present invention;

[0047] Figure 7 A schematic diagram of adding a planar map with a fixed depth according to an embodiment of the present invention;

[0048] Figure 8 A schematic diagram of an underwater three-dimensional reconstruction result obtained by an embodiment of the present invention;

[0049] Figure 9 A flowchart of three-dimensional reconstruction on a numerical measurement line according to an embodiment of the present invention;

[0050] Figure 10Schematic diagram of sonar image;

[0051] Figure 11 Spatial schematic diagram expanded from sonar image;

[0052] Figure 12 Schematic diagram of underwater three-dimensional reconstruction device based on UUV cruise and two-dimensional sonar.

[0053] Marking description: 1. Bridge pier, 2. Depth plane d1, 3. Robot equipped with this set of three-dimensional program system, 4. Depth plane d2, 5. Map of depth plane d1, 6. Map of depth plane d2, 7. Diving route Specific implementation manner

[0054] The following further elaborates on the specific implementation manner of the present invention in conjunction with the accompanying drawings.

[0055] An underwater three-dimensional reconstruction method based on UUV cruise and two-dimensional sonar provided by the present invention includes:

[0056] S1: First, turn on the depth-keeping function of the robot 3 and drive the robot 3 to the target depth d 1 , in this mode, the robot moves on the depth plane 2 of this d 1 ;

[0057] S2: Fuse the dead reckoning information and the forward-looking sonar information to realize the map construction on the depth plane d 1 , and obtain the depth plane d 1 ; The method for constructing the plane map is the two-dimensional cartographer algorithm, using the dead reckoning result on this depth plane as the odometer input and the leading point cloud information of the forward-looking sonar image as the point cloud input; Creating a two-dimensional map of a certain depth plane has the advantages of being able to initially explore the underwater environment and providing position calibration for subsequent underwater three-dimensional reconstruction. Figure 5

[0058] The dead reckoning information comes from the data fusion of DVL, IMU, and depth gauge, that is, the three-axis velocity obtained by DVL, the three-axis attitude, angular velocity, acceleration obtained by IMU, and the depth obtained by the depth gauge, and the current position and attitude of the robot are deduced through the extended Kalman filter.

[0059] The forward-looking sonar information is the occupancy probability of the spatial point cloud converted from the forward-looking sonar image. Substantially, any point (R, θ) on the sonar image corresponds to a small spherical arc in space with a distance of [R, R + ΔR] and an azimuth angle interval of [θ, θ + Δθ], and the pitch angle The probability of an object appearing in a small spherical arc is positively correlated with the pixel value I(R, θ) of this point. The specific conversion method is as follows:

[0060] S2.1: Obtain an image with length and width of sonar R sonar respectively from a sonar with an action range radius of R sonar , an azimuth angle of θ and a vertical aperture of φ pixel . Here, R pixel represents the number of pixels of the sonar imaging radius on the image. Convert it into a spatial voxel with length, width and height numbers of R pixel and respectively. For the convenience of representation, each voxel is represented by a point cloud with uniform distribution, and the origin of the sonar imaging is used as the origin of the spatial coordinate system. The default value of each point is -1, indicating unknown. As shown in Figure 10 and Figure 11 , obtain an image with length and width of R respectively from a sonar with an action range radius of R, an azimuth angle of θ and a vertical aperture of pixel R pixel . Here, R represents the number of pixels of the sonar imaging radius on the image. Convert it into a spatial voxel with length, width and height numbers of pixel R and

[0061] respectively.

[0062]

[0063] S2.2: Traverse each point cloud (X, Y, Z) and convert it to the polar coordinate system pixel . For the points satisfying R ∈ [0, R , which is the actual imaging area corresponding to this frame of sonar image, obtain the occupancy probability of this voxel from the sonar image through the corresponding R and θ.

[0064] S2.3: Normalize the occupancy probability of the point cloud (X, Y, Z), that is

[0065]

[0066] I max is the maximum pixel value in the sonar image, 255 for an 8-bit image, I min is the minimum pixel value, generally 0. The range of P(X, Y, Z) is (0 - 1). The larger the value, the greater the probability that the voxel space corresponding to this point cloud is occupied.

[0067] S3: Change the depth of the depth control to d 2 , repeat the steps of S2, and obtain the ground 2 of the depth plane at d Figure 6, and through prior conditions, match the x-y coordinates of these two maps;

[0068] As Figure 1 , Figure 2 and Figure 3 shown, in the embodiment of the present invention, after constructing the maps on the d1 plane and the d2 plane through the Cartographer algorithm, during the process of x-y coordinate matching, according to the known information in terms of position, "the cross-section of each depth of the pier structure is axisymmetric and centrosymmetric", so the geometric center P1 of the image in the d1 plane map and the geometric center P2 of the d2 plane map have the same x and y coordinates in the space coordinates, and only differ in the depth z; in terms of angle, according to the known information that "the cross-section of each depth of the pier structure has a length greater than the width in the length direction", rotate the d2 plane map to align with the d1 plane map after the same short side direction.

[0069] As Figure 2 and 3 shown, it is an embodiment of the present invention. The robot carrying the three-dimensional reconstruction system of the present invention constructs two-dimensional maps of the respective depth planes on the d1 depth plane and the d2 depth plane. The map constructed on the d1 plane is as Figure 4 shown, and the map constructed on the d2 plane is as Figure 5 shown.

[0070] S4: The robot 3 performs preliminary positioning in the map of the d 1 depth plane, and after turning on the control of the x azimuth, y azimuth, and heading angle, it performs vertical diving scanning, that is, only moves along the z-axis until it reaches the d Figure 5 depth plane 4; 2

[0071] Figure 6 As shown, it is an embodiment of the present invention. After the robot constructs two plane maps, it turns on the control of x, y, and heading, and only dives and floats in the depth direction to scan the underwater structure.

[0072] Since the diving process of the robot from the d 1 depth plane 2 to the d 2 depth plane 4 is relatively fast, the offset error in the x and y directions is relatively small, and it usually drifts in a certain direction. Therefore, according to the new positioning coordinates in the map of the d 2 depth plane, linearly correct the diving path; record the matching coordinates of the robot in the map of the d 1 depth plane as (x Figure 6 1 , y Figure 5 ), and the robot in the d 1 , y 2 ) 2 depth plane of the mapFigure 6 The coordinates of 2 are (x 2 ), then at depth d i (d 1 < d i < d 2 ), the position coordinates (x(d i ), y(d i )) can be approximated as

[0073]

[0074] And according to the need, the number of fixed-depth plane maps can be increased to improve the imaging accuracy.

[0075] S5: Re-position in the ground 2 at depth d. The difference between the coordinates in the ground Figure 6 at depth d 2 and the coordinates in the ground Figure 6 at depth d 1 is the offset in the diving route 7. Based on this offset, a linear correction is made to the diving route 7 of the robot in S4; as Figure 5 shown in Figure 7 , in an embodiment of the present invention, in order to obtain higher 3D reconstruction accuracy, more fixed-depth plane maps are used to calibrate the path deviation of the robot during the diving process.

[0076] S6: Based on the corrected path information, splice and filter the forward sonar information, and then the spatial point cloud information scanned along this path can be obtained;

[0077] For the splicing and filtering of the forward sonar information, the idea of synthetic aperture and spatial carving algorithms is adopted. For each voxel, the minimum value in all observations is maintained. The specific steps are as follows:

[0078] S6.1 Add the point P new generated by the latest frame to the environmental point cloud P e , and update the value of the original environmental point cloud P e

[0079]

[0080] S6.2: After completing this survey line, process the global point cloud on this survey line, set an intensity threshold E threshold , and remove the points in the point cloud of this survey line that are less than this energy threshold, that is, remove the uncertain points and low-probability points. The remaining points are the 3D imaging results on this survey line.

[0081] S7: As needed, replace the vertical scanning path with a rising path if necessary, and repeat the steps of S4 - S6. Select to switch the survey line on the depth plane where the target depth plane map has been constructed. Based on the target depth plane map, enhance the positioning information of the ROV during the survey line switching. Thus, obtain the spatial point cloud information scanned along different paths, perform rough registration according to the pose, and then perform fine registration of the point cloud based on the ICP method to obtain the overall three-dimensional point cloud reconstruction of the underwater structure. As Figure 8 shown, it is the three-dimensional reconstruction effect finally obtained in an embodiment of the present invention. As Figure 9 shown, it is the three-dimensional reconstruction flowchart of an embodiment of the present invention on a numerical survey line. First, the UUV moves along the vertical direction and performs sonar imaging. The forward-looking sonar information is the occupancy probability of the spatial point cloud converted from the forward-looking sonar image, and it is stitched with the environmental point cloud of this survey line that has been constructed, and the environmental point cloud is updated. If the three-dimensional reconstruction of this survey line has been completed, perform overall filtering to remove uncertain points and low-probability points.

[0082] The depth-keeping function described in step S1 and the control of the x azimuth, y azimuth, and heading angle described in step S3 adopt PID control; the feedback value of the depth-keeping control comes from the depth gauge; the navigation angle value of the heading control comes from the IMU; the position values in the x and y directions come from dead reckoning.

[0083] The beam of the forward-looking sonar has a certain elevation divergence characteristic. When the surface of the underwater structure is not perpendicular to the imaging plane or there are obvious abrupt parts of the underwater structure entering the sonar imaging area, after the initial imaging, the UUV can be made to scan and image the target structure close by to flexibly improve the three-dimensional reconstruction accuracy.

[0084] As Figure 1 shown, corresponding to the foregoing method embodiment, there is also provided an underwater three-dimensional reconstruction system based on UUV cruise and two-dimensional sonar. The system includes three parts: a UUV carrier platform, an environmental perception module, and a data processing module. The UUV carrier platform includes a motion controller, an actuator, and a mechanical structure. The environmental perception module includes a motion perception module and an acoustic perception module: the motion perception module includes sensors related to motion perception, such as an IMU, a depth gauge, and a DVL; the acoustic perception module is an acoustic perception sensor, including a forward-looking sonar and a sounding sonar. The environmental perception module is carried by the UUV carrier platform, and the data processing module receives the data of the environmental perception module, runs a three-dimensional reconstruction algorithm, and issues control commands to the UUV carrier platform.

[0085] As Figure 1-8The modules used in the embodiments of the present invention are as follows: The system is deployed on an ROV platform. The forward-looking sonar is the Oculus M750d of BluePrint. This sonar has a low price and a small volume, making it suitable for being carried by small and medium-sized UUVs. Although the beam opening angle of nearly 20° will cause a decrease in accuracy, when deployed on small and medium-sized UUVs, combined with a pan-tilt head, the imaging angle can be adjusted to achieve a large-range imaging of the seabed projection, which is more practical. The bathymetric sonar is a single-beam sonar, the ISA500 of Impact Subsea in the UK, which is used to prevent the ROV from colliding with the seabed during the diving and scanning process. In the motion perception module, the IMU selects the BW-AH300, the depth gauge is the KELLER Series 10LX, and the DVL is the WaterLinked A50. The data of the environmental perception module is finally transmitted to the dry-end ground station through power line carrier, and a three-dimensional reconstruction algorithm is run on a local PC.

[0086] Using this set of solutions, experiments are carried out in an environment with relatively small fluctuations. The accuracy of the two-dimensional map of the fixed-depth plane created by the present invention can reach more than 95%, and the accuracy of the overall underwater three-dimensional reconstruction is more than 85%.

[0087] The forward-looking multi-beam sonar can also be replaced with a multi-beam profile sonar. Compared with the M750d, its characteristic is a small beam elevation angle and higher accuracy in the elevation direction, enabling more accurate underwater three-dimensional reconstruction.

[0088] Corresponding to the foregoing embodiments of an underwater three-dimensional reconstruction method based on UUV cruising and two-dimensional sonar, the present invention also provides an embodiment of an underwater three-dimensional reconstruction device based on UUV cruising and two-dimensional sonar.

[0089] See Figure 12 , an underwater three-dimensional reconstruction device based on UUV cruising and two-dimensional sonar provided by the embodiments of the present invention includes a memory and one or more processors. An executable code is stored in the memory. When the processor executes the executable code, it is used to implement an underwater three-dimensional reconstruction method based on UUV cruising and two-dimensional sonar in the above embodiments.

[0090] The embodiments of an underwater three-dimensional reconstruction device based on UUV cruising and two-dimensional sonar provided by the present invention can be applied to any device with data processing capabilities. The any device with data processing capabilities can be a device or apparatus such as a computer. The device embodiments can be implemented through software, or through hardware or a combination of software and hardware. Taking software implementation as an example, as a logically meaningful device, it is formed by the processor of any device with data processing capabilities reading the corresponding computer program instructions in the non-volatile memory into the memory and running them. From the hardware level, such as Figure 12As shown, it is a hardware structure diagram of any device with data processing capabilities where the underwater three-dimensional reconstruction device based on UUV cruise and two-dimensional sonar provided by the present invention is located. Except for Figure 12 the processor, memory, network interface, and non-volatile memory shown, in an embodiment, any device with data processing capabilities where the device is located usually includes other hardware according to the actual functions of the any device with data processing capabilities, which will not be elaborated here.

[0091] The implementation processes of the functions and roles of each unit in the above device are specifically described in detail in the implementation processes of the corresponding steps in the above method, which will not be elaborated here.

[0092] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can be referred to the partial description of the method embodiment. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of the present invention. Those of ordinary skill in the art can understand and implement it without creative work.

[0093] The embodiment of the present invention also provides a computer-readable storage medium, on which a program is stored. When the program is executed by a processor, it implements an underwater three-dimensional reconstruction method based on UUV cruise and two-dimensional sonar in the above embodiment.

[0094] The computer-readable storage medium can be an internal storage unit of any device with data processing capabilities described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be an external storage device of any device with data processing capabilities, such as a plug-in hard disk, a Smart Media Card (SMC), an SD card, a Flash Card, etc. equipped on the device. Further, the computer-readable storage medium can also include both an internal storage unit and an external storage device of any device with data processing capabilities. The computer-readable storage medium is used to store the computer program and other programs and data required by the any device with data processing capabilities, and can also be used to temporarily store the data that has been output or will be output.

[0095] The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the underwater three-dimensional reconstruction method based on UUV cruise and two-dimensional sonar.

[0096] Other embodiments of the present application will be readily apparent to those skilled in the art upon consideration of the specification and practice of the disclosure herein. The present application is intended to cover any variations, uses, or adaptations of the present application, which follow the general principles of the present application and include known common general knowledge or conventional technical means in the technical field not disclosed in the present application. The specification and examples are only illustrative, and the true scope and spirit of the present application are pointed out by the claims.

[0097] It should be understood that the above general description and the following detailed description are exemplary and explanatory only and should not limit the present application. The present application is not limited to the exact structures already described and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present application is only limited by the appended claims.

Claims

1. An underwater 3D reconstruction method based on UUV cruising and 2D image sonar, characterized in that: The method includes: S1. Obtain a first target depth and a second target depth according to a height range of the space to be reconstructed, and lower the robot to determine the first target depth and the second target depth; S2. Use the robot to collect dead reckoning information and forward-looking sonar information, and use the obtained information to construct a map of the target depth to obtain a first target depth plane map and a second target depth plane map; S3, moving the robot to the first target depth for positioning, and moving and scanning in the vertical direction; when the robot moves to the second target depth, performing secondary positioning, and correcting the movement route according to the coordinate difference in the map; S4, based on the corrected path information, stitching and filtering the point cloud in the forward-looking sonar information to obtain spatial point cloud information; S5. Acquire the spatial point cloud information scanned by different paths for registration to obtain the overall three-dimensional point cloud reconstruction of the underwater structure.

2. The underwater three-dimensional reconstruction method based on UUV cruising and two-dimensional image sonar according to claim 1 is characterized in that: The dead reckoning information is used to calculate the current position and posture of the robot, specifically including: three-axis speed obtained by DVL, three-axis posture obtained by IMU, angular velocity, acceleration and depth obtained by depth meter.

3. The underwater three-dimensional reconstruction method based on UUV cruising and two-dimensional image sonar according to claim 1 is characterized in that: The forward-looking sonar information is the spatial point cloud occupancy probability converted from the forward-looking sonar image, and the conversion process is specifically as follows: The length and width obtained by sonar are R pixel The image is converted into length, width and height respectively. R pixel , The spatial voxel of pixel It is expressed as the number of pixels on the image within the sonar imaging radius. Each voxel is represented by a uniformly distributed point cloud. The origin of the sonar imaging is used as the origin of the spatial coordinate system. The default value of each point is -1, indicating unknown. Traverse each point cloud (X, Y, Z) and convert it to polar coordinate system For R∈[0,R pixel ]、 The point is the actual imaging area corresponding to the sonar image of this frame. The occupancy probability of the voxel is obtained from the sonar image through the corresponding R and θ, where θ sonar is the azimuth of the sonar, sonar is the vertical aperture of the sonar The point cloud (X, Y, Z) occupancy probability is normalized, that is, I max is the maximum pixel value in the sonar image, I min is the minimum pixel value, the range of P(X,Y,Z) is (0~1), and the larger the value is, the greater the probability that the voxel space corresponding to the point cloud is occupied.

4. The underwater three-dimensional reconstruction method based on UUV cruising and two-dimensional image sonar according to claim 1 is characterized in that: The target depth plane map construction includes: using the dead reckoning result on the depth plane as the odometer input, the front point cloud information of the forward-looking sonar image as the point cloud input, and using a two-dimensional cartographer algorithm to complete the construction.

5. The underwater three-dimensional reconstruction method based on UUV cruising and two-dimensional image sonar according to claim 1 is characterized in that: The correction of the motion route includes: the matching coordinates of the robot in the first target depth plane map are (x1, y1), the coordinates of the robot in the second target depth plane map are (x2, y2), and the coordinates of the robot in the depth plane map are (x2, y2). i The position coordinates (x(d i ), y(d i ))Corrected to where d1 <d i <d2。 6. The underwater three-dimensional reconstruction method based on UUV cruising and two-dimensional image sonar according to claim 1 is characterized in that: The stitching and filtering of the point cloud in the forward-looking sonar information includes: for each voxel, keeping the minimum value among all observations, the specific steps are as follows: S6.1 The point P generated by the latest frame new Add to the environment point cloud P e And update the original environment point cloud P e Value S6.2: After completing the survey line, process the global point cloud on the survey line and set an intensity threshold E threshold , remove the points whose energy is less than the energy threshold in the point cloud of the survey line, that is, remove the uncertain points and low-probability points, and the remaining points are the three-dimensional imaging results on the survey line.

7. The underwater three-dimensional reconstruction method based on UUV cruising and two-dimensional image sonar according to claim 1 is characterized in that: The spatial point cloud information scanned by different paths specifically includes: selecting a depth plane on which a target depth plane map is constructed to switch the survey line, and enhancing the positioning information of the ROV when switching the survey line based on the target depth plane map.

8. A system for implementing the method according to any one of claims 1 to 7, characterized in that: The system includes: UUV carrier platform, environment perception module and data processing module; The UUV carrier platform is an ROV or AUV, which is used to undertake movement and control functions and is equipped with an environmental perception module to collect dead reckoning information and forward-looking sonar information; after receiving the collected data, the data processing module realizes three-dimensional reconstruction of the underwater scene; The environmental perception module includes a motion perception module and an acoustic perception module; the motion perception module is an IMU, a DVL, and a depth meter; the acoustic perception module is a forward-looking sonar and a height-finding sonar.

9. The system according to claim 8, characterized in that The UUV carrier platform includes a mechanical body, an actuator and a motion controller; the mechanical body includes a frame and buoyancy material; the actuator includes a thruster and a gimbal; the motion controller is used to receive control instructions issued by a data processing module to complete the control of the actuator.

10. An underwater three-dimensional reconstruction device based on UUV cruising and two-dimensional image sonar, comprising a memory and one or more processors, wherein the memory stores executable code, characterized in that: When the processor executes the executable code, an underwater three-dimensional reconstruction method based on UUV cruising and two-dimensional image sonar is implemented as described in any one of claims 1-7.

Images